I've been trying to build a desktop application for the entire pandemic. In my professional work I work on infrastructure products (think software engineering, not DevOps or SRE). My skills are not really outfit for frontend development or desktop development. Shocking stuff, my ineptitude, I know (/sarcasm)
I ended up learning some JS and I wrote a desktop app with the help of a package called Wails for Go. It's electron-like in many ways, but lets me code up a "backend" in go while JS is resigned to doing things it's good at, like UI. Eventually, though, I struck the gold mine. I actually made a native desktop app in Go with an immediate mode package. It looked like shit, but it worked. I then went on to design a command line variant as well.
My chief takeaway after doing all of them: even my native code resembled the patterns found in JavaScript based UI developments. This basically implies "ELM is everywhere" to some degree, even in Frontends that don't explicitly say it. QT's QML is JavaScript that's pre-compiled for desktop. Sciter is the same way. You can gripe about electrons size, computational, and memory efficiency, but the opposite (native programming for UI) requires an order of magnitude of very niche programming that will only be useful for this purpose, of which any cool thing (like syntax highlighting) that you might want to do will need to be done for the first time. Just a quick reminder, again, I'm not really a fan of or regular user of JavaScript or TypeScript.
What I'm saying is, there are tradeoffs to be made. I don't blame a bunch of startups investing in a JavaScript desktop app when you can absolutely produce a stable cash cow with minimal skillset change and leave a handful of teams building independent threads and daemons that JavaScript can pass work off to that it can't do.